RFR: 8306965: osThread allocation failures should not abort the VM

[David Holmes]

Currently OSThread allocation will abort the VM if there is not enough memory. While the fact we are exhausting C-heap means we are likely to abort for one reason or another, it isn't the role of thread starting to act as that trigger. So `new OSThread()` becomes `new (std::nothrow) OSThread()` and the existing null checks actually serve a purpose.

Change was suggested by Kim Barrett - thanks.

Testing:
 - tiers 1-3 sanity
 - GHA sanity

Thanks

-------------

Commit messages:
 - 8306965: osThread allocation failures should not abort the VM

Changes: https://git.openjdk.org/jdk/pull/13879/files
 Webrev: https://webrevs.openjdk.org/?repo=jdk&pr=13879&range=00
  Issue: https://bugs.openjdk.org/browse/JDK-8306965
  Stats: 8 lines in 4 files changed: 0 ins; 0 del; 8 mod
  Patch: https://git.openjdk.org/jdk/pull/13879.diff
  Fetch: git fetch https://git.openjdk.org/jdk.git pull/13879/head:pull/13879

PR: https://git.openjdk.org/jdk/pull/13879